We all know somebody, or somebody who knows sombody, who had a winning streak in the casino, and then proceeded to lose it all because he overstayed his welcome, so to speak. Or how about the trader who had a winning streak and doubled up on his trades, until his last one didn’t go as planned and brought down the bank. We are all suffering the consequences of revered economists who looked at the data and ignored the ones that didn’t conform to their pre-conceived construct. You wonder: are these people reckless? Are they addicted to the adrenaline rush of winning? The answer is that they are as rational as you and me. They made the all too human mistake of extrapolation. We tend to think that if things went in a certain direction, they will continue to do that. Of course, intellectually we know this not to be true, but when it comes to our behavior in real life situations we behave as if it is true. Why do we do it?
The selective advantage of extrapolation
Don’t forget, our brain evolved and adapted in an environment that was radically different from today’s environment. Things were simpler and change was slower. If we found grains growing in the open savannah it would make sense to deduce that we’d find grains in another open field. If we found plenty of animals to hunt during the migration of the wilderbeast after the rainy season, wouldn’t it make sense to assume that the following year would yield the same bountiful protein so essential to our survival? In fact, it makes so much natural-selection sense that any individual who fails to make this deduction is bound to become extinct. In fact, this capacity to make deductions is so important that it must be hard-wired in the brain.
But things have changed since those simple times. We now have to make decisions in a rapidly changing world, and a much more complex one. Can we safely make the deduction that if we ran a red light numerous times without bad consequences that next one is not going to end badly? Of course not. Yet many of us keep running the red light, assuming that past performance predicts future outcome.
The wall street casino
I was intrigued by Jason Zweig’s Friday column in the Wall street Journal; today’s column is titled The intelligent investor: this is your brain on a hot streak. There he examines the subject of predicting future events based on past performance, and how we all fall prey to it. Jason Zweig also wrote a delightful book, Your Money and Your Brain,that transcends the narrow subject of money; it delves into how we make decisions, many of them fatefully wrong. It’s a must read; it occupies a place of honor on my nightstand.
Decision Research, a nonprofit think tank in Eugene, Ore., has conducted a nationwide online survey of investors seven times since 2008.
These surveys have shown that investors’ forecasts of future returns go up after the market has risen and down after it has fallen.
William Burns, an analyst at Decision Research, says investors’ forecasts of the market’s return over the coming year were heavily swayed by how stocks performed in the previous month.
We are hard-wired to do it
The assumption that making deductions must be etched in our brain because it has selective advantage makes sense. but is there hard evidence?
Paul Glimcher and Hannah Bayer of New York University Center for Neural Sciences published a brilliant paper on the subject. In essence what they showed is that the reward system in our midbrain has specialized cells that calculate a sort of moving average of past events, giving the greatest weight to the most recent outcomes. What’s a moving average? Let say that on day 1 we perform a certain task and get reward A. On day 2 we perform the same task and get a larger reward B, on day 3 the same task will yield a yet larger reward C, and so on. Now let’s assume that these cells have the capacity to contain only 7 days’ worth of information. So on day 8, they drop the information of day 1, on day 9 they drop the informtion of day 2, and so on. If these cells integrate (or average) all the stored information, what you get is a “moving average”. The fact that older information is constantly dropped to make room for the most recent one means that newer information has more weight in the moving average.
Now comes the fascinating observation: if the newly-acquired information is in agreement with the direction of the previous information (for instance, increased reward for a given task) the cells will release the neurotransmitter dopamine. This substance is the major neurotransmitter of the brain’s reward system, the very same system that is responsible for the pleasure of sex, and dope, and a good massage. Is there any surprise that we like making extrapolations, especially if they are of a streak of success? We readily attribute it to our wisdom, which makes us feel good. Attributing it to dumb luck is not going to cut it.
Which immediately raises the question: how do the cells behave if one day negative information arrives? Do they subtract the event from the moving average? Do we conclude that the winnig streak was a fluke and is now over? Not really. The cells simply don’t release the dopamine, and will resume the release when new information reaches them which conforms to the previous trend. In other words, inconvenient truths are simply ignored.
Daniel Kahnemann, Nobel Prize winner for his work on behavioral economics, loves to tell this story:
When he was a young officer in the Israeli Defense Force he was assigned the task of evaluating the leadersip qualities of young officer-cadets. He and his psychologist colleagues devised a task for the cadets to perform as a group, and then observed who assumed leadership, who showed creativity, etc. Their military superiors were highly pleased by the scientific selection method. Being budding scientists, the young psychologists decided to evaluate the validity of their evaluations. What they found was that some cadets who had received high marks in their tests indeed developed into good leaders ,but many didn’t. Their selection method did not differ from pure random selection They wrote a report and submitted it to headquarters. The bosses adamantly refused to believe the study: after all, they could point to many officers who were correctly selected by this test methodology. Why would they deny the facts? Kahnemann and his colleague Tversky studied that as well; the officers simply ignored the negative data points, accepting only the ones that confirmed their beliefs.
There is more to the story than miltary obtuseness; trained scientists fall into the same trap, making excuses to explain “bad data”. There must be an independent system in the brain that deals with negative information, with things that “don’t fit” and make us less satisfied with ourselves. Otherwise, how are we going to learn from experience? There is evidence that such a system does exist, probably in the brain stem, most likely involving serotoninergic neurons.
So next time your dopamine-addled brain flatters your intelligence, don’t fall for it; you are simply extrapolating from past events, and selectively at that. Sooner or later your streak will come to an end.